Sepsis is a common, complex, life-threatening syndrome arising from serious infection. Early, tailored therapy is crucial to the outcome of sepsis, but reliable endpoints for therapy have proved elusive. This lack of reliable Statistical and signal-processing analyses of variability in the body's cardiovascular control system may prove useful in defining valid endpoints for early therapy of sepsis, as well as identifying patients at high risk for poor outcome from sepsis. Employing the highly detailed electronic medical record at Intermountain Healthcare, this proposal aims to study patterns in the variability of heart rate and arterial blood pressure in a large cohort of patients with severe sepsis and septic shock. This research hypothesizes that spectral and non-linear measures of cardiovascular variability like spectral power, sample entropy, and fractal exponents will predict recovery of cardiovascular homeostasis by 24 hours after presentation in septic patients. This research further hypothesizes that changes in cardiovascular variability in response to early therapies will improve our ability to predict recovery of cardiovascular homeostasis. Finally, this research hypothesizes that patterns in variability will predict immediate response to administration of vasopressor drugs or intravenous fluids. This research intends to lay the groundwork for improved bedside care of patients with sepsis as well as better understanding of mechanisms of disruption and restoration of homeostasis in patients with sepsis.